R Notebook

# install.packages("keras")
# library(keras)
# install_keras()

Libraries

library(quantmod)
library(tseries)
library(timeSeries)
library(forecast)
library(xts)
library(dplyr)

df <- read.csv("./data/bitcoin_coinbase_M5.csv", sep = "\t", header = TRUE)
df2 <- read.csv("./data/btc-out.csv", sep = "\t", header = TRUE)
head(df2)

df$DateTime_UTC <- as.POSIXct(df$DateTime_UTC)
df2$DateTime_UTC <- as.POSIXct(df2$DateTime_UTC)
# quantiles ==== [(-42.016, -0.185], (-0.185, -0.0749], (-0.0749, -0.0298], (-0.0298, -0.00504], (-0.00504, 0.0], (0.0, 0.0131], (0.0131, 0.0394], (0.0394, 0.085], (0.085, 0.19], (0.19, 67.889]
head(df)

data <- df2
N <- nrow(data)
plot(data$price_close, type = "l")

plot(diff(data$price_close), type = "l", main = "Original data")

plot(diff(log(ms)), type = "l", main = "Log-transformed data")

Error in diff(log(ms)) : object 'ms' not found

make_aic_table <- function(data, P_end, Q_end, P_start = 0, Q_start = 0){
  table <- matrix(NA,(P_end-P_start+1),(Q_end-Q_start+1))
  for(p in P_start:P_end) {
    for(q in Q_start:Q_end) {
      cat(paste0("\rP: ", p, " and Q: ", q))
       table[p-P_start+1,q-Q_start+1] <- arima(data,order=c(p,0,q))$aic
    }
  }
  dimnames(table) <- list(paste("<b> AR",P_start:P_end, "</b>", sep=""),paste("MA",Q_start:Q_end,sep=""))
  table
}

table(aic_table(ms_df,40,40),digits=2)

spfinal.aic <- Inf
spfinal.order <- c(0,0,0)
for (i in 0:4) for (j in 0:4) {
  print(paste0("Testing i: ", i, " and j: ", j))
  spcurrent.aic <- AIC(arima(ms_df, order=c(i, 0, j)))
  if (spcurrent.aic < spfinal.aic) {
    spfinal.aic <- spcurrent.aic
    spfinal.order <- c(i, 0, j)
  }
}

spfinal.order

[1] 4 0 4

make_aic_table(ms_df, 7, 7, 4, 4)

P: 4 and Q: 4
P: 4 and Q: 5

m1_arima <- arima(price_close, order=c(30, 0, 30))

Error in optim(init[mask], armafn, method = optim.method, hessian = TRUE,  : 
  non-finite finite-difference value [10]

m2_arima <- arima(price_close, 7, 0, 7)

# Pull data from Yahoo finance 
getSymbols('TECHM.NS', from='2012-01-01', to='2015-01-01')

# Select the relevant close price series
stock_prices = TECHM.NS[,4]
In the next step, we compute the logarithmic returns of the stock as we want the ARIMA model to forecast the log returns and not the stock price. We also plot the log return series using the plot function.

# Compute the log returns for the stock
stock = diff(log(stock_prices),lag=1)
stock = stock[!is.na(stock)]

# Plot log returns 
plot(stock,type='l', main='log returns plot')

adf.test(y, alternative = "stationary")

Error in NCOL(x) : object 'y' not found